Four areas of theoretical population genetics will be studied. In the first, we will develop mathematical and statistical framework for studying the evolution of genes that affect how other sets of genes interact with one another. This theory helps in understanding conditions under which the genome can be expected to become modular in its production of phenotypes, as well as when synergy between mutations in different genes should be produced by evolution. We shall study how the pattern of interaction between deleterious mutations evolves as a function of the mutation rate, the recombination rate, and the degree of fitness loss of each mutation. We shall also study buildup of statistical association between genes that influence culturally transmitted traits that are associated either through transmission or through fitness. The second area of study concerns the evolution of pathogens such as influenza. Here we build models to predict the accumulation of mutations during epidemics and pandemics in a way that can assist in guiding vaccination strategies. We will develop algorithms that search for potential recombinants and reassortants among a collection of up to 1000 viral sequences. Here we aim to devise a statistical test to indicate whether identified recombinants or reassortants are spurious or statistically significant. Our recent theory of niche construction will form the basis of studies of how pathogens might evolve in response to human actions that they induce, such as use of antibiotics. The final research area will develop multiple-gene models for sex-linked control of genomic imprinting. These will include cis and trans modifiers of imprinting. Fertility selection as well as sex- specific viabilities will be studied in order to clarify the role of multiple paternities, which has been proposed as a driving force in the evolution of genomic imprinting. Genomic data from mammalian species will be analyzed using tools from statistical learning in order to predict which genes are likely to be imprinted. Correlations between DMA sequence properties and predicted imprinting status based on life history characteristics will also be sought. [unreadable] [unreadable] [unreadable]